Moving beyond the hype of Generative AI: It’s power and pitfalls Free webinar from AI leaders at AI Fund, Google, and Albertsons... Listen Now x

Nicolas Känzig

Machine Learning Engineer

A skilled machine learning engineer and data scientist, Nicolas brings years of experience working with startups to bear in his role advising portfolio companies. He helps them research and develop their ML tech and evaluate what is possible.

Nicolas comes to AI Fund from the Colombian unicorn Rappi, where he was a data scientist focusing on an AI-enabled marketing platform as well as ML models for automated data enrichment and data cleaning. Prior to that, he worked as a data scientist and machine learning engineer at several other startups in Colombia and Switzerland and led ML workshops for industry and university audiences. 

Originally hailing from Switzerland and now living in Colombia, Nicolas speaks four languages and is working on his fifth—Portuguese.

He also particularly enjoys playing musical instruments, having grown up in a musical family and learning to play guitar, bass, piano, and percussion instruments. “Music for me is meditation,” he says. “It’s a way to express yourself and after a long day of programming, it’s great to get the guitar out and really dive into the music.”

He holds a master’s degree in information technology and electrical engineering and bachelor’s degree in the same, both from ETH Zurich.

A trend in AI that Nicolas is especially excited about putting into action is transfer learning and its evolution:

“In the past, people would train a new model from scratch for each task, but methods such as transfer learning have enabled models to make use of previously gained knowledge. However, this comes with a big restriction: the new task has to be related closely to the previous task the model has been trained on,” he says.

“The next generation of this takes transfer learning a step further. AI architectures that acquire a more general understanding of the world and can be applied to a variety of different tasks. Imagine models that can both solve computer vision and natural language processing tasks, while transferring the knowledge gained on one domain to the other,” continues Nicolas.


Ready to build something great?